1

How Do Exporters Adjust to Exchange-Rate Fluctuations? New Evidence from the East African Community

Alan Asprilla, Univerity of Lausanne

Nicolas Berman

Graduate Institute of International Studies, Geneva and CEPR

Olivier Cadot University of Lausanne, CEPR and FERDI

Marguerite Duponchel

International Growth Center

Mélise Jaud The World Bank

UNU-WIDER conference, Learning to Compete, Helsinki, June 24-25 2013

2

KEY POLICY QUESTIONS

EAC pursuing two-pronged regional integration strategy o Trade integration

Customs union Attempts at cooperating on building a common market through Reductions in NTBs MRAs for some types of services

o Monetary integration Before embarking into monetary integration, we need to understand i. How exporters adapt to exchange-rate fluctuations (exchange-rate pass-through) ii. What is the real cost of exchange-rate volatility on trade Our strategy: Use our answer to (i) to infer extent of market power (lack of trade integration) in EAC.

3

DO MONETARY UNIONS GROW FASTER?

0.00

0.50

1.00

1.50

2.00

2.50

3.00

3.50

4.00

WAEMU EAC CEMAC

Average annual growth rate ofreal GDP per capita, 2000-2011

o Monetary unions, like fixed exchange-rate zones, are vulnerable to asymmetric shocks

o Lack of market integration raises the probability of asymmetric shocks, so market integration and monetary integration are linked

o Oil exploitation in some of EAC’s member states (Uganda) will be a major asymmetric shock

o There is little prima-facie evidence that Africa’s monetary unions have grown faster than other zones

4

Freund Pierola (2012) on export surges:

Obstfeld Rogoff (2002) on the exchange-rate disconnect

EXCHANGE-RATE POLICY IS KEY TO EXPORT GROWTH

5

…AND SO IS EAC’S REGIONAL MARKET, WHICH BREEDS A SPECIAL TYPE OF FIRMS—SMALL MANUFACTURERS

And the most regionally specialized are the smallest

Close to 60% of EAC’s exporters realize over

95% of their export turnover on regional

markets

6

EXCHANGE-RATE PASS-THROUGH: WHAT THEY DO, WHAT WE DO

Our dependent variable: Producer price in LCU

7

Country level estimates Feenstra (1989): ERPT into U.S. prices around 0.6; i.e. if exchange-rate doubles (from say €0.7/USD to €1.4/USD), U.S. consumer price goes down by only 30% on average across studies Marston (1990): Even less ERPT (0.1-0.5, PTM 0.5-0.9), variable across sectors o Incomplete ERPT—pricing to market—taken as evidence of variable

markups (with constant markups, ERPT would be 100%), imperfect competition, market segmentation

Firm-level estimates Surprisingly consistent PTM estimates (around 0.1, implying ERPT around -0.9) across countries (Atkeson and Burstein (2008), Berman et al. (2012), Fosse (2012), Chaterjee et al. (2012) o More PTM for large firms, more PTM for core products, more PTM for

more productive firms

WHAT THE LITERATURE SAYS… EVIDENCE FROM PTM

8

TESTABLE COMPARATIVE-STATICS PROPERTIES WITH ADDITIVE DISTRIBUTION COSTS

From the theory (standard model as in Berman et al. 2012, Chatterjee et al. 2012) : Prices: 1- More productive firms price more to market 2- More pricing to market in destinations with higher distribution costs 3- Less pricing to market in faraway destinations 4- Less pricing to market in destinations where competition is tougher Volumes: 5- More productive firms have lower volume elasticity 6- Lower volume elasticity in destinations with higher distribution costs 7- Higher volume elasticity for faraway destinations 8- Higher volume elasticity in destinations where competition is tougher

9

NOTATION

10

IDENTIFICATION STRATEGY

Estimation issues 1. Exchange-rate exogenous to pricing—no endogeneity bias here 2. Firm size approximated by number of products endogenous to exchange

rate—we’ve got a problem here Instrumentation & excuses

Lag number of products—not terribly powerful Define number of products at firm level, not firm-product-definition

PTM coefficient βp

11

DATA: THE FIRST MULTI-COUNTRY FIRM-LEVEL DATASET

Export transaction data from customs administrations of 6 developing countries

o The good: Large sample o The bad: No firm-level covariates except constructed from the database o The ugly: very, very noisy data, especially when it comes to unit values

12

PTM: BASELINE RESULTS

Dependent var.: ln (Unit Value)Estimator: OLS

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)

Log bilateral RER 0.108*** 0.0853** 1.622*** -0.0812 -0.0908 -0.197*** 0.0873*** 0.137*** 0.0695** 0.0692 -0.559 -0.0225(0.0316) (0.0332) (0.369) (0.127) (0.212) (0.0692) (0.0317) (0.0303) (0.0309) (0.390) (0.370) (0.352)

Interaction terms

ln (RER) × deval. a/ -0.00217 0.000232 0.000670 0.000608(0.00143) (0.00136) (0.00136) (0.00136)

ln (RER) × ln (dist.) -0.182*** -0.0612 0.0490 -0.0385(0.0439) (0.0430) (0.0434) (0.0397)

ln (RER) × ln (dest. GDP/cap) 0.0223* -0.0141 -0.000750 -0.00824(0.0128) (0.0252) (0.0238) (0.0237)

ln (RER) × ln (dest. GDP) 0.00987 0.0167 0.0145 0.0249*(0.00779) (0.0144) (0.0133) (0.0131)

ln (RER) × manuf. Prod. 0.396*** 0.301*** -0.122** -0.106*(0.0777) (0.0707) (0.0572) (0.0568)

ln (RER) × ln (1+number prod.) b/ 0.00848*** 0.00588***(0.00211) (0.00203)

ln (RER) × ln (lag number prod.) b 0.00570*** 0.00413** 0.00449**(0.00194) (0.00192) (0.00192)

ln (RER) × EAC bilateral trade c/ 0.692*** 0.341** 0.525***(0.153) (0.164) (0.179)

Devaluation (Real) 0.0155*** 0.0104** 0.00671 0.00691(0.00495) (0.00491) (0.00477) (0.00477)

ln (dest. GDP/cap) -0.190*** 0.546*** 0.476*** 0.515***(0.0480) (0.0999) (0.104) (0.103)

ln (dest. GDP) -0.323*** -0.648*** -0.505*** -0.539***(0.0476) (0.0897) (0.0921) (0.0912)

ln (1+number prod.) 0.00230 0.00749(0.00677) (0.00672)

ln (lag number prod.) -0.0103 -0.00688 -0.00746(0.00646) (0.00644) (0.00644)

Observations 568,275 568,275 568,275 567,172 567,114 568,240 568,275 431,635 568,275 566,990 430,556 430,556R-squared 0.967 0.967 0.967 0.967 0.967 0.967 0.967 0.969 0.967 0.967 0.969 0.969Firm-product-destination FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes YesOrigin--year FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes

13

VOLUME ELASTICITIES

Dependent var.: ln (Volume)Estimator: OLS

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)

Log bilateral RER 0.403*** 0.514*** 0.380 2.220*** 3.094*** -0.0612 0.402*** 0.469*** 0.438*** 3.629*** 3.035*** 2.324***(0.0655) (0.0710) (0.589) (0.276) (0.441) (0.123) (0.0658) (0.0749) (0.0666) (0.811) (0.866) (0.789)

Interaction terms

ln (RER) × deval. a/ -0.00247 -0.00286 0.000885 0.000966(0.00282) (0.00285) (0.00294) (0.00294)

ln (RER) × ln (dist.) 0.00270 -0.193** -0.0344 0.0816(0.0699) (0.0917) (0.102) (0.0840)

ln (RER) × ln (dest. GDP/cap) -0.202*** 0.0192 0.0317 0.0416(0.0274) (0.0530) (0.0550) (0.0549)

ln (RER) × ln (dest. GDP) -0.109*** -0.0897*** -0.122*** -0.136***(0.0163) (0.0316) (0.0327) (0.0320)

ln (RER) × manuf. Prod. 0.601*** 0.682*** 0.674*** 0.652***(0.133) (0.134) (0.142) (0.141)

ln (RER) × ln (1+number prod.) b/ 0.00142 0.00415(0.00385) (0.00383)

ln (RER) × ln (lag number prod.) b -0.0120*** -0.00529 -0.00578(0.00359) (0.00362) (0.00361)

ln (RER) × EAC bilateral trade c/ -0.633*** -0.813*** -0.696*(0.227) (0.291) (0.360)

Devaluation (Real) -0.0470*** -0.0514*** -0.0540*** -0.0543***(0.0106) (0.0107) (0.0108) (0.0108)

ln (dest. GDP/cap) 1.015*** -0.615*** -0.644** -0.697***(0.113) (0.230) (0.250) (0.250)

ln (dest. GDP) 1.024*** 1.544*** 1.687*** 1.733***(0.100) (0.199) (0.216) (0.215)

ln (1+number prod.) 0.250*** 0.244***(0.0129) (0.0128)

ln (lag number prod.) 0.0587*** 0.0427*** 0.0435***(0.0122) (0.0122) (0.0122)

Observations 568,278 568,278 568,278 567,175 567,117 568,243 568,278 431,637 568,278 566,993 430,558 430,558R-squared 0.931 0.931 0.931 0.931 0.931 0.932 0.932 0.934 0.931 0.932 0.934 0.934Firm-product-destination FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes YesOrigin--year FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes

14

PTM, EAC EXPORTERS

Dependent var.: ln (Unit Value)Estimator: OLS

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)

Log bilateral RER -0.103 -0.127 2.865*** 0.134 0.662 -0.568*** -0.131 0.125 -0.314*** -0.749 -1.534 0.0102(0.106) (0.110) (0.670) (0.334) (0.596) (0.142) (0.107) (0.0926) (0.115) (1.340) (1.087) (0.929)

Interaction terms

ln (RER) × deval. a/ -0.00571* -0.000875 -0.000164 -0.00106(0.00327) (0.00323) (0.00322) (0.00320)

ln (RER) × ln (dist.) -0.378*** 0.197 0.281** -0.0155(0.0851) (0.161) (0.140) (0.0910)

ln (RER) × ln (dest. GDP/cap) -0.01000 0.0412 0.105 0.102(0.0358) (0.0815) (0.0676) (0.0676)

ln (RER) × ln (dest. GDP) -0.0205 -0.0624 -0.0621 -0.0242(0.0227) (0.0505) (0.0420) (0.0401)

ln (RER) × manuf. Prod. 0.925*** 0.645*** 0.124 0.0743(0.178) (0.177) (0.162) (0.161)

ln (RER) × ln (1+number prod.) b/ 0.0114** 0.00531(0.00466) (0.00443)

ln (RER) × ln (lag number prod.) b 0.00532 0.00287 0.00263(0.00361) (0.00364) (0.00365)

ln (RER) × EAC bilateral trade c/ 0.862*** 0.725** 0.888***(0.188) (0.327) (0.312)

Devaluation (Real) 0.0170 0.00322 -0.000842 -0.000687(0.0127) (0.0127) (0.0121) (0.0121)

ln (dest. GDP/cap) -0.705*** 0.624* 0.0328 0.0517(0.166) (0.361) (0.290) (0.290)

ln (dest. GDP) -0.874*** -1.116*** -0.574** -0.589**(0.148) (0.295) (0.248) (0.248)

ln (1+number prod.) 0.0181 0.0223(0.0161) (0.0161)

ln (lag number prod.) -0.0134 -0.0103 -0.0106(0.0132) (0.0133) (0.0133)

Observations 145,181 145,181 145,181 144,872 144,873 145,181 145,181 112,501 145,181 144,801 112,189 112,189R-squared 0.957 0.957 0.957 0.957 0.957 0.957 0.957 0.962 0.957 0.957 0.962 0.962Firm-product-destination FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes YesOrigin--year FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes

15

SUMMING UP RESULTS

Whole sample o PTM coefficient around 0.1 without all the interaction terms

Like in the rest of the literature—no difference between industrial and developing countries?

More ERPT at the firm level (0.9) than aggregate/sector-level ERPT (0.3 on average)

o Volume elasticities very high for the whole sample—when doing the algebra, assuming 20% transportation (τ) cost and 100% retail margin (η), estimates imply elasticity of substitution (σ) between 4 and 8

EAC exporters o In general, no PTM for EAC exporters, implying no market power o But very strong PTM (0.7 < βp < 0.9) on EAC markets (bilateral trade), suggesting

substantial market power o Weak supply response, suggesting binding capacity constraints

16

EFFECT OF EXCHANGE-RATE VOLATILITY ON ENTRY AND EXIT

Dependent var.:

Sample EAC bilateral All Sample EAC bilateral All SampleEstimator: RE Probit (1) (2) (3) (4)

RER volatility a/ 4.088*** 5.199*** -0.619 -0.511**(1.320) (0.536) (0.522) (0.211)

Financial dependence b/ -0.186* -0.191*** -0.0482 0.0240(0.0984) (0.0551) (0.0370) (0.0195)

Volatility × Financial dependence -1.710 1.618* 1.096 -0.543(1.831) (0.858) (0.813) (0.372)

ln (distance) -2.147*** -0.436*** -0.0784** 0.0981***(0.174) (0.0215) (0.0357) (0.00493)

ln (dest. GDP/cap) 0.194 -0.360*** 0.201* 0.0402***(0.469) (0.0198) (0.112) (0.00472)

ln (dest. GDP) 0.289*** 0.301*** -0.0316* -0.0185***(0.0784) (0.0131) (0.0179) (0.00295)

Firm scope c/ -0.530*** -0.401*** -0.00408 0.00231(0.0215) (0.0117) (0.00377) (0.00218)

Fixed effectsFirm-product-destination Yes Yes Yes YesOrigin-year Yes Yes Yes Yes

Observations 42,751 122,735 89,217 243,155Number of Firm-Destination-Product cells 29,072 81,699 47,101 138,453

Entry Exit

17

CONCLUSIONS

Pricing to market behavior of exporters suggests strong evidence of market power on EAC markets: o Markets still segmented, protected by tariffs (25% band), NTBs o Difficult arbitrage between infant-industry protection and need to discipline abuses of

market power Entry and exit behaviour does not provide strong evidence of damage from exchange-rate volatility: o Exit rates go down with exchange rate volatility o Exit rates not higher for credit-constrained firms Policy implications o Focus on pursuing regional trade integration (good compromise between IIP and

opening) o Still looking for compelling case to launch process of monetary integration (given

tremendous costs in terms of macro constraints)

Thank you!

18

19

VOLUME ELASTICITY TO THE EXCHANGE RATE

In partial equilibrium (P and Y constant), log-differentiating x(φ) w.r.t. e gives

20

THE FUNDAMENTAL PRICING TO MARKET EQUATION

21

THE FUNDAMENTAL PRICING TO MARKET EQUATION

22

KEY DEFINITIONS

PTM

ERPT

That is, with a 10% PTM coefficient (βp = 0.1), a 100% exchange-rate increase (from 2’000 shillings per dollar to 4’000) translates into a 10% rise in the shilling producer price and a 90% drop in the dollar consumer price.

Slide Number 1Slide Number 2Slide Number 3Slide Number 4Slide Number 5Slide Number 6Slide Number 7Slide Number 8Slide Number 9Slide Number 10Slide Number 11Slide Number 12Slide Number 13Slide Number 14Slide Number 15Slide Number 16Slide Number 17Slide Number 18Slide Number 19Slide Number 20Slide Number 21Slide Number 22